sign in sign up
<<
Home , Arctic Alaska, Maar, Seward Peninsula

VOL. 49, NO. 1 (MARCH 1996) P. 62Ð69

The Largest Known on , , Northwest JAMES E. BEGÉT,1 DAVID M. HOPKINS1 and STEVEN D. CHARRON2

(Received 23 January 1995; accepted in revised form 5 October 1995)

ABSTRACT. The Espenberg Maars on the northern Seward Peninsula of Alaska were formed by a series of basaltic eruptions through thick . The maars were excavated as much as 300 m into older lithologies; ranging from 4 to 8 km in diameter, they are the four largest known maars on earth. Hydromagmatic eruptions which derive water from ground ice are evidently extremely explosive. The high heat capacity of ice in permafrost modulates the supply of water interacting with during the eruption, producing consistently low coolant-to-fuel ratios in an environment with a sustained, abundant water supply. The Espenberg Maars demonstrate that, under certain conditions, eruptions which involve the interaction of and permafrost are powerful enough to produce craters as large as small . Key words: , permafrost, arctic, Alaska, hydromagmatic

RÉSUMÉ. Les maars de l’Espenberg situés dans la partie septentrionale de la péninsule Seward en Alaska ont été formés par une série d’éruptions basaltiques datant du pléistocène, à travers une forte épaisseur de pergélisol. Les maars ont été creusés à une profondeur allant jusqu’à 300 m dans d’anciennes roches; d’un diamètre variant entre 4 et 8 km, ils sont les quatre plus grands maars connus sur Terre. Les éruptions hydromagmatiques qui tirent l’eau de la glace de sol sont, comme on l’a déjà constaté, extrêmement explosives. La grande capacité thermique de la glace dans le pergélisol détermine l’approvisionnement en eau qui interagit avec le magma au cours de l’éruption, donnant régulièrement lieu à un faible rapport refroidissant / combustible dans un environnement où l’eau est constamment abondante. Les maars de l’Espenberg démontrent que, dans certaines conditions, les éruptions qui déclenchent une interaction lave-pergélisol sont suffisamment puissantes pour donner naissance à des cratères de la grandeur de petites calderas. Mots clés: maar, pergélisol, arctique, Alaska, hydromagmatique

Traduit pour la revue Arctic par Nésida Loyer.

INTRODUCTION Although maars are the second most common volcanic landform after cinder cones (Cas and Wright, 1987), no Maars are shallow but broad craters formed by explosive previous examples have been documented of maars produced excavation into older lithologies during phreatomagmatic by eruptions through permafrost. Here we describe the eruptions. The Espenberg Maars are found on the northern- geomorphology and model the thermodynamic processes most Seward Peninsula (Fig. 1), just south of the , that occur during the interaction of magma and permafrost. and farther north than any other Late Pleistocene volcanoes in Such interactions led to highly explosive eruptions and the (Wood and Kienle, 1990). Because of their formation of unusually large maars on the Seward Peninsula high-latitude setting in Alaska, these maar eruptions occurred of northwest Alaska. in an area where permafrost is ca. 100 m thick. The presence of basaltic rocks on the northern Seward Peninsula was first noted by Moffitt (1905). Hopkins (1959, DIMENSIONS OF HYDROVOLCANIC CRATERS 1963) identified extensive Quaternary basaltic lava flows, AND MAAR especially in the Imuruk area of the central Seward Peninsula, and later described the Late Quaternary Espenberg The Espenberg Maars consist of four large, separate erup- Maars and coeval tephra layers (Hopkins,1988). Radiocar- tive craters, each excavated 100Ð300 m into Pleistocene bon dates and tephrochronology, together with relative age sediments and . Most of the maars are circular to slightly estimates based on the degree of and sedimentation in elliptical in shape, except for the Devil Mountain Maar, the lake basins, indicate that the Devil Mountain Maar is the which is an irregular composite crater containing North and youngest, formed ca. 17 500 years B.P. (Table 1); South South Devil Mountain Lakes (Fig. 1). North Devil Mountain Killeak Maar is > 40 000 years old; North Killeak Maar is Lake (5.1 km in diameter) is partly separated from South somewhat older; and Whitefish Maar may be 100Ð200 000 Devil Mountain Lake (3.4 km diameter) by a small sand spit. years old (Hopkins, 1988; Begét et al., 1991). Sequences of bedded surge deposits, airfall lapilli beds,

1 Department of Geology and Geophysics, University of Alaska, Fairbanks, Alaska 99775-0760, U.S.A. 2 Department of Geology and Geography, University of Massachusetts, Amherst, Massachusetts 01003, U.S.A. © The Arctic Institute of North America SEWARD PENINSULA MAARS ¥ 63

Chukchi Kitluk River Kalik DevilDevil 66¡30' River MountainMountain Maar Espenberg River

Singeakpuk River North North Killeak Goodhope Killeak Bay Maar Whitefish Maar South N Thick KilleakSouth Killeak tephra Maar Lava flows and cones Kongachuk 0 5 10 164¡30' Creek 164¡00' 66¡15' km

FIG. 1. Location of the Espenberg Maars on the northern Seward Peninsula, Alaska, showing Devil Mountain Maar, Whitefish Maar, North Killeak Maar, and South Killeak Maar. Surrounding each maar is a thick blanket of pyroclastic surge and airfall tephra deposits; the pattern shows the distribution of tephra more than 5 m thick surrounding each maar.

TABLE 1. Radiocarbon dates on organic material buried by Devil Peninsula produced by the eruptions of the Devil Mountain Mountain Lakes Maar tephra, Seward Peninsula, Alaska. Lakes Maar all indicate an age of about 17 500 years B.P. (Table 1). The existence of only a single tephra deposit at Radiocarbon Laboratory Material Location numerous sites around the maar, the similarity of all radiocar- Date Number Dated bon dates from beneath the tephra deposits, and the absence 16 990 ± 150 B-18632 peat (macrofossil) Nuglungnugtuk River of significant stratigraphic breaks in either the distal ash or 17 420 ± 260 B-60718 peat (macrofossil) Lake Rhonda the proximal volcaniclastic sequence indicate the Devil Moun- 17 630 ± 800 W-3489 peat (macrofossil) Whitefish Thaw Pond 17 740 ± 110 B-75731 peat (macrofossil) Lumpy Drained Lake tain Lakes Maar formed during one complex eruptive episode 17 740 ± 220 B-75529 peat (macrofossil) Tempest Lake about 17 500 years B.P. The crater produced during those 17 910 ± 1501 B-18549 detrital twigs Kiliwooligoruk Creek eruptions, measured from rim to rim across the maar lake, is 16 880 ± 120 B-60716 wood (macrofossil) Egg Lake 17 980 ± 100 B-71983 sediment Egg Lake 8 km long by 6 km wide, as much as 200 m deep, and covers over 30 km2 (Fig. 1). 1 in reworked alluvium The maar craters at North and South Killeak Lakes, lo- cated 20 km east of the Devil Mountain Maar, are 4.0 and 5.0 massive pyroclastic flows, and explosion breccia can be km in diameter and ca. 20 and 12 km2 in area, respectively traced through many fresh exposures in 10Ð40 m high cliffs (Fig. 1). Whitefish Maar lies 15 km west of the Devil around the Devil Mountain Lakes Maar (Fig. 2a, 2b). Mountain Maar, is 4.3 km in diameter, and covers 15 km2, Stratigraphic sections were measured through the volcanic although the lake basin has been partly filled by alluvium and pile at many sites around the lakes, and complex but uninter- other sediments since the eruption. Exposures through proxi- rupted sequences of plane-bedded and cross-bedded surge mal hydromagmatic deposits are rare at the older maars, but deposits, massive explosion breccia, and scoria beds were isolated exposures exist in stream gullies at each older maar, found. No trace of nonvolcanic sediment or paleosol develop- and in each case the absence of paleosols, nonvolcanic ment was found at any level within the sequence of sediment, or other stratigraphic breaks suggests each of these volcaniclastic sediments (Fig. 2b). Similarly, radiocarbon maars formed as the result of complex but monogenetic dates from organic material preserved at numerous sites eruptive events. beneath a widespread tephra deposit on the northern Seward In comparison to the Espenberg Maars of the Seward 64 ¥ J.E. BEGÉT et al.

FIG. 2. a) Exposure of volcaniclastic sediments near the southwest shore of North Devil Mountain Lake, showing plane-bedded surges overlying massive pyroclastic flows and explosion breccia. The outcrop is 45 m high. b) Outcrop of explosion breccia near the eastern shore of South Devil Mountain Lake, containing numerous clasts of unconsolidated sands, peats, and other materials. The dumbell-shaped clast in the lower center of the photograph consists of unconsolidated, bedded medium sand, thought to have been frozen when emplaced. The knife (base of photograph) is 2 cm across.

Peninsula, virtually all maars at lower latitudes are much observed elsewhere. These craters evidently mark the sites of smaller. Average diameters of typical maars are only a few explosive activity during the eruption; other explosion cra- hundred to a thousand meters, and cover less than 1 km2 in ters may have formed earlier in the eruption sequence, only to area (Cas and Wright, 1987). The largest such features be obliterated or filled in by later eruptions. Although we are previously described are about 3 km in diameter, with surface unable to find any published bathymetric data from other areas less than 8 km2 (Fig. 3). The Espenberg Maars are maars, we speculate that multiple craters may exist on other therefore several times larger in diameter than other maars, maar lake floors. Similar groups of coeval explosion craters and an order of magnitude larger in surface area. have recently been documented at sites like Cerro Xalapaxco The underwater topography of several of the Alaskan maar in central , where multiple phreatomagmatic erup- lakes was studied by depth profiling during the summer of tions occurred at a site with an especially abundant water 1992. The lake basins in Devil Mountain Lakes Maar are supply (Abrams and Siebe, 1994). locally more than 100 m deep (Fig. 4) and, if we include the height of bedrock cliffs around the lake, the maar craters have been excavated as much as 250 m into older sediments and HYDROMAGMATIC ERUPTIVE PROCESSES lava flows. The maximum water depth at North Killeak Lake AT THE ESPENBERG MAARS is about 25 m, while South Killeak Lake is more than 60 m deep; but these lakes lie in the bottom of craters excavated The Espenberg Maars are the first reported example of 200Ð 250 m into older lava flows. Whitefish Lake was hydromagmatic eruptions produced by interactions of magma not profiled, but is much shallower than the other lakes, with permafrost (Table 2). The well-documented 1977 consistent with its greater age and degree of infilling. Unkinrek Maar eruption in (Kienle et al., The bottom of the Devil Mountain Lakes Maar is exten- 1980; Self et al., 1980) occurred in an area 1500 km south of sively cratered, with at least eight separate depressions vis- the Espenberg Maars, and outside the southern limits of ible in the contour data (Fig. 4). These well-defined craters permafrost in Alaska (Péwé, 1975). Field studies at other range from 0.1Ð1 km in diameter and are 50 Ð100 m deep. high-latitude maars, such as the 1875 Viti Maar in The underwater craters seem to trend north-south in South (Sigurdsson and Sparks, 1978), also found no evidence for Devil Mountain Lake, but east-west in North Devil Mountain permafrost playing a role in the eruption. Lake, paralleling the ellipitical asymmetry of each lake basin. The processes involved in the interaction of permafrost Similar but partly infilled craters are also visible in depth and magma at the Espenberg Maars are more than a curiosity, profiles from the somewhat older Killeak Lakes. as the unusually large size of these four maars indicates that We interpret the closed depressions on the floors of the hydromagmatic eruptions involving permafrost can be sig- Espenberg maars as small explosion craters, as they are nificantly more explosive and typically excavate craters similar in size to typical, small hydromagmatic craters larger than those resulting from interactions with surface or SEWARD PENINSULA MAARS ¥ 65

30

5 20 N = 121 25'

10 ¡ 66

10 Number

90 10 5 80

100 5

NK W SK DM 50 0 0 1 2 3 4 5 6 7 8 9 Diameter (km)

FIG. 3. Histogram of diameters of known maars, showing typical diameters are 1Ð2 km. The four Espenberg maars (Devil Mountain = DM, North Killeak = NK, South Killeak = SK, and Whitefish = W) are significantly larger than any previously described maar, and constitute the four largest-known maars on 110 North earth. Figure modified from Cas and Wright (1987).

70 (Table 2). Massive ground ice and ice wedges 0 0.5 1.0

occur in surface exposures around the Seward Peninsula 80 Kilometers today, and disseminated ice is present throughout the frozen sediments and within the fractured basaltic flows which form the substrate in the area of the Espenberg 164¡35' 10 Maars. Data from several drill holes on the northern Seward Peninsula indicate that the ground is currently FIG. 4. Bathymetry of Devil Mountain Maar, showing the lake floor. The maar is locally more than 100 m deep. Small (0.2Ð1.0 km diameter) depressions frozen to depths of as much as 100 m in this area; perma- delineated by concentric, circular topographic lines may be individual craters frost was probably somewhat thicker during the Pleistocene, formed by phreatomagmatic explosions late in the eruption. These craters when the eruptions occurred (Péwé, 1975). record excavation of these large maars during numerous hydromagmatic The processes of interaction between lava and water are explosive episodes at shallow depths. known to be complex (Sheridan and Wohletz, 1981, 1983). TABLE 2. Water sources for selected hydromagmatic eruptions However, the physical and thermodynamic properties of ice that have produced maars and cones. differ considerably from those of liquid water, and introduce additional steps and complexity. The thermal conductivity of Water sources Example Crater Excavation ice is several orders of magnitude lower than that of lava, so diameter (m) Depth (m) ice in direct contact with lava can insulate deeper ice and Maars: restrict melting to the surfaces actually in contact. This is permafrost Espenberg Maars, Alaska 4000Ð8000 300 illustrated by the observation that lava flows which traverse ground water Ukinrek Maars, Alaska 300 ca. 100 generally do not cause rapid melting of large vol- shallow lake Taal, 200 ca. 50 umes of snow and ice (Major and Newhall, 1989). Tuff Cones: The strong contrast between the heat capacity of basaltic shallow marine Diamond Head, Hawaii 1700 0 -1 -1 shallow lake Pavant Butte, Utah 460 10 lava, estimated at ca. 0.3 cal gm deg (McBirney, 1993), and playa Cerro Colorado, Mexico 0 0 that of ice and frozen soil, at 2.3Ð 4.0 cal gm-1 deg-1 (Williams and Smith, 1989), will also tend to inhibit rapid melting of ground ice, as 10 grams of lava must drop a degree in temp- steam and phreatic explosions can occur. Altogether, these erature to produce each degree rise in temperature of a gram thermodynamic constraints dictate that about two-to-three of frozen sediments. Pleistocene permafrost temperatures at times more energy is needed to melt ice in permafrost and the Espenberg Maars were probably ca. -7û to -40ûC, depend- bring the meltwater to the boiling point than would be ing on depth, weather, and the time of year. In addition, the required to boil an equivalent mass of groundwater or surface latent heat of fusion of ice is 80 cal gm-1, but the latent heat water beginning at ambient surface temperatures. released as lava solidifies produces only ca. 60 cal gm-1 These physical constraints clearly indicate that ground ice (McBirney, 1993), further tending to minimize melting dur- which comes into contact with lava melts and flashes to steam ing lava-ice interactions. Finally, even after some ground ice considerably more slowly than an equivalent amount of has melted, an additional 100 cal must be extracted from the liquid groundwater. Also, the total amount of liquid water lava to heat each gram of water to 100ûC before it can flash to produced by melting of ground ice during an eruption is 66 ¥ J.E. BEGÉT et al. continuously limited by the thermodynamic properties of both permafrost and lava. These constraints occur throughout Kilauea the course of an eruption, even though abundant ice is present Miles N 0 in the frozen sediments. 1 Sheridan and Wohletz (1981, 1983) showed that the 0 123 explosivity of hydromagmatic eruptions is maximized when Kilometers water-to-magma ratios are low. Experimental tests suggest that fuel-coolant reactions are most explosive when thermal energy is efficiently transmitted to the coolant, and water and magma mixtures become well mixed and approach thermal equilibrium most completely at ratios of ca. 0.3Ð0.5 (Wohletz, 1986). At most maars, highly explosive conditions constitute a transient eruptive phase associated either with initially low supplies of groundwater, or with the exhaustion of available North water supplies during the course of the eruption. Killeak In contrast, the intrinsic difficulties of melting ground ice Devil Lake during eruptions through permafrost would tend to produce Mountain the low water-to-magma ratios which are associated with Lakes high explosivity, while the ubiquitous supply of ice in the frozen sediments would ensure that water is continually South available throughout the course of the eruption. The total Killeak amount of water generated by melting at any time would most Lake likely be small in relation to the volume of lava in contact with permafrost. Low water-to-magma ratios could theoretically be maintained throughout the course of even prolonged eruptions. Hydromagmatic eruptions through permafrost are therefore likely to be characterized by sustained, highly FIG. 5. Devil Mountain Maar, and North and South Killeak Maars, when explosive conditions. plotted at the same scale as Kilauea Caldera in Hawaii, show that the Espenberg maars are similar in size or larger than small volcanic calderas. Scarps and The conditions of abundant total water supply and limited concentric fractures around the Alaskan maars (shown by concentric lines) water availability during an eruption are usually mutually record landslides and other mass movements in the surrounding permafrost, exclusive. However, eruptions in areas of continuous perma- and indicate that these maars grew in part as the result of slope failure and frost, like that of the ca.100 m thick permafrost layer on the slumping into the maar craters. northern Seward Peninsula, may encounter an abundant supply of ice—present throughout the entire area affected by grew from a combination of several processes, including the eruptions, bonded into all surface sediments, soil, and rocks, coalescence of several vents during extended eruptions and and locally exceeding 50% by volume—while at the same repeated collapse of landslide debris from the margins of the time the supply of liquid water is limited by thermodynamics. crater. The landslides contributed to the mixing of blocks of Lorenz (1986) suggested that maars grow as crater walls frozen ground with magma, and to the highly explosive collapse into successively deeper and wider conduits. The character of the eruptions. shapes of North and South Devil Mountain Lakes and the Some other unusual eruptive processes may have been evidence of multiple craters on the floors of the lakes (Fig. 4) important at these high-latitude maars. For instance, the indicate that complex vent structures, perhaps controlled by depth of excavation of the maars is similar to but slightly fissures, existed during these arctic eruptions, and influenced greater than the modern thickness of permafrost. Perhaps this the shape of the maar craters. However, these craters also reflects ponding of magma or water that produced explosions grew as a consequence of landsliding into the vents, as at the base of the permafrost during the eruptions. Also, landslide complexes, some more than 1 km wide, ring the significant quantities of methane and methane hydrate some- Espenberg Maars today (Fig. 5), and clearly played an impor- times accumulate under permafrost (Kvenvolden, 1988; tant role in the growth of these large maars (Hopkins, 1988). Nisbet, 1989); it is conceivable that methane explosions These landslide complexes are delineated today by 1Ð2 m occurred in the course of the eruptions at the Espenberg high scarps which cut across topographic features, as well as Maars. by parallel stream channels and by elongated lakes. Volcaniclastic sediments surrounding Devil Mountain Landslides of high-enthalpy, water-rich debris into vents Maar and the other Espenberg Maars contain evidence of can produce hydromagmatic explosions (Houghton and Nairn, numerous surges and explosions during these hydromagmatic 1991), and since permafrost is extremely prone to failure eruptions. Sequences of plane-bedded and cross-bedded surge during thawing, it is likely that mixtures of frozen debris and deposits, Strombolian airfall beds, and explosion breccias are high-enthalpy water and steam were introduced into vents by exposed in 10Ð80 m thick sections which completely sur- landslides at the Espenberg Maars. These maars therefore round the lakes (Figs. 2a, 2b). The surge deposits consist of SEWARD PENINSULA MAARS ¥ 67

blocky, lithic fragments in a sand-sized -lithic matrix. smaller craters during complex but monogenetic eruptions. Accretionary lapilli are uncommon, suggesting water con- While the Espenberg maars are currently the only known tents of the surges were low. Base surges may have swept examples of caldera-sized maars formed by permafrostÐ across the wide platforms, visible in the bathymetric data, that magma interactions, it is important to note that broadly surround the craters in the center of the maar lakes (Fig. 4). similar processes may operate during hydromagmatic erup- At least two layers of explosion breccia, each 1Ð5 m tions at maars at lower latitudes. For instance, Wohletz and thick, are exposed in bluffs on the southwest shoreline of Heiken (1992:92) suggested that caldera structures at the North Devil Mountain Lake, and similar breccias are Baccano caldera in Italy “appear to represent the coalescence exposed at other sites around the lakes (Fig. 2a). The clasts of several maar vents that eventually collapsed together to in the explosion breccia deposits at Devil Mountain Maar form a single caldera.” The Baccano caldera is about 3 km in consist of rounded boulders of layered soft clays, silts, diameter (Funiciello et al., 1979). sand, bedded coarse sand and pebbles, and peats and Lorenz (1986) suggested that maars form when partially petrified wood, as well as occasional massive hydromagmatic explosions “ream out” funnel-shaped vents basalt clasts derived from older basalt flows (Hopkins, that had become backfilled with ejecta and wall-collapse 1988). The many fragile blocks of soft sediment, ranging material. His model closely resembles the purely magmatic from 0.5 to 2 m in diameter (Fig. 2a), must have been process hypothesized to have formed some Japanese frozen when transported and emplaced. These blocks pro- calderas (Yokoyama, 1958, 1960), albeit on a smaller scale. vide direct evidence of the explosive interaction of magma The large craters of the Espenberg Maars are not simple, and permafrost during these eruptions. single vents, but evidently formed when multiple smaller craters coalesced. The presence of permafrost apparently tends to produce this type of eruption, as each of the four FORMATION OF CALDERA-SIZED CRATERS Espenberg Maars of records a sustained, highly BY HYDROMAGMATIC ERUPTIONS explosive, hydromagmatic eruption which produced a caldera- THROUGH PERMAFROST sized crater. Although similar maars have not been described from A caldera is defined as a “large volcanic , more other areas of the world, it seems likely that other examples or less circular or cirque-like in form, the diameter of which of unusually large maar craters exist. Such large maar craters, is many times greater than that of the included vent or vents, formed by interaction of permafrost and magma, would no matter what the steepness of the walls or form of the floor” necessarily be restricted to high-latitude areas in the Arctic or (Bates and Jackson, 1980:89; cf. Williams, 1941). A diameter , or to high-elevation areas like the , of at least 2Ð3 km is usually taken as the minimum size for or to mid-latitude areas where Pleistocene permafrost tempo- a caldera. No single causal mechanism is dictated by this rarily existed. definition, which is found in virtually all contemporary volcanology and geomorphology textbooks, as well as in geologic dictionaries and glossaries published by geologic SUMMARY AND CONCLUSIONS societies. It is occasionally suggested that the term caldera be restricted only to large collapse craters formed by ignimbrite Devil Mountain Lakes Maar on the northern Seward eruption, but in the strict sense this is incorrect and inconsist- Peninsula is the largest known maar on earth, and South ent with general usage in the geologic and volcanologic Killeak, North Killeak, and Whitefish Maars are also larger communities. For example, well-defined calderas exist at than any previously described maars on earth. The Espenberg Kilauea and Mauna Loa (Carr and Greeley, 1980), at sites Maars were formed by highly explosive hydromagmatic clearly associated with effusive rather than explosive ignim- eruptions through permafrost. The explosivity of brite eruptions. Reviews by Williams (1941), Yokoyama (1960), hydromagmatic eruptions, like other fuel-coolant reactions, Smith and Bailey (1968), Williams and McBirney (1979), is quite sensitive to the ratio of the components. The thermo- and Walker (1984) all discuss a variety of caldera-forming dynamic properties of ice, water, and lava constrain the rate processes together with representative examples of calderas. at which water can be produced by melting. The low thermal As shown in this paper, hydromagmatic eruptions through conductivity, high heat capacity, and high latent heat of permafrost on the northern Seward Peninsula of Alaska have fusion of ground ice would tend to produce low water-to- repeatedly excavated caldera-sized craters, forming the larg- magma ratios during eruptions through ground ice, leading to est known maars on earth (Begét and Mann, 1992). The Devil highly explosive conditions. The ubiquitous presence of Mountain Lakes Maar is similar in diameter to the classic thick ground ice in the perennially frozen ground would Crater Lake caldera in (Williams, 1941), and all of the provide a continuous source of water throughout the course of Espenberg Maars are larger than well-known small calderas, even a prolonged eruption. such as the summit calderas at Kilauea and Mauna Loa in The four Espenberg Maars are similar in size to or larger Hawaii or Katmai in Alaska (Fig. 5). The four than small calderas like those at Kilauea, Katmai, and Crater Espenberg craters are therefore examples of caldera-sized Lake. Multiple small craters on the lake floors, revealed by maars, formed by explosive excavation and coalescence of bathymetric surveys, indicate that the maars were formed by 68 ¥ J.E. BEGÉT et al. the coalescence of multiple vents. Landsliding of frozen ———. 1988. The Espenberg Maars: A record of explosive volcanic sediment from the margins of the craters played an important activity in the Devil Mountain-Cape Espenberg area, Seward role in the growth of these unusually large maars. Radiocar- Peninsula, Alaska. In: Schaaf, J., ed. The Bering Land Bridge: An bon dating and stratigraphic studies suggest that these maar archeological survey. Nome, Alaska: U.S. Park Service. 188Ð247. craters formed during a series of complex but monogenetic HOUGHTON, B.F., and NAIRN, I.A. 1991. The 1976Ð1982 eruptions. Hydromagmatic eruptions through permafrost can, Strombolian and phreatomagmatic eruptions of White Island, in certain cases, produce caldera-sized craters by shallow, : Eruptive and depositional mechanisms at a “wet” explosive excavation of older lithologies. volcano. Bulletin of Volcanology 54:25Ð49. KIENLE, J., KYLE, P.R., SELF, S., MOTYKA, R.J., and LORENZ, V. 1980. Ukinrek Maars, Alaska. I. April 1977 eruption sequence, ACKNOWLEDGEMENTS petrology, and tectonic setting. Journal of Volcanology and Geothermal Research 7:11Ð 37. We are grateful to the Bering Land Bridge National Park and the KVENVOLDEN, K.A. 1988. Methane hydrates and global climate. National Park Service for providing financial and logistic support Global Biogeochemical Cycles 2:221Ð229. for our work at the Espenberg Maars. Support for Begét was also LORENZ, V. 1986. On the formation of maars and diatremes and received from the Division of Polar Programs of the National its relevance to the formation of tuff rings. Bulletin of Volcanology Science Foundation and the Alaska Volcano Observatory, a joint 48:265Ð274. operation of the U.S. Geological Survey, the University of Alaska, MAJOR, J., and NEWHALL, C. 1989. Snow and ice perturbation and the Division of Geological and Geophysical Surveys of the during historical volcanic eruptions and the formation of lahars State of Alaska. Dr. Kenneth Wohletz reviewed an early version of and floods. Bulletin of Volcanology 52:1Ð27. this manuscript. Patricia Craw, Stephen Moffatt, and Tim McBIRNEY, A.R. 1993. Igneous petrology. Second edition. Boston: Tannenbaum ably provided assistance to the field parties, and Dr. Jones and Bartlett. 508 p. Mary Edwards, Dr. Dan Mann, Vickie Goethchus and Claudi MOFFITT, F.H. 1905. The Fairhaven gold placers, Seward Hoefle made helpful suggestions during the fieldwork. Buck Maxon Peninsula, Alaska. U.S. Geological Survey Bulletin 247. 203 p. of Arctic Air Guides provided air taxi service from Kotzebue to the NISBET, E.G. 1989. Some northern sources of atmospheric methane: Espenberg Maar lakes and back. Production, history, and future implications. Canadian Journal of Earth Sciences 26:1603Ð1611. PÉWÉ, T.L. 1975. Quaternary . U.S. Geological REFERENCES Survey Professional Paper 835. 207 p. SELF, S., KIENLE, J., and HUOT, J.P. 1980. Ukinrek Maars, ABRAMS, M.J., and SIEBE, C. 1994. Cerro Xalapaxco: An unusual Alaska. II. Deposits and formation of the 1977 craters. Journal tuff cone with multiple explosion craters, in central Mexico of Volcanology and Geothermal Research 7:39Ð65. (Puebla). Journal of Volcanology and Geothermal Research SHERIDAN, M.F., and WOHLETZ, K.H. 1981. Hydrovolcanic 63:183Ð199. exposions I. The systematics of water-pyroclast equilibration. BATES, R.L., and JACKSON, J.A. 1980. Glossary of geology. Science 212:1387Ð1389. Second edition. Falls Church, Virginia: American Geological SHERIDAN, M.F., and WOHLETZ, K H. 1983. Hydrovolcanism: Institute. 751 p. Basic considerations and review. Journal of Volcanology and BEGÉT, J., and MANN, D. 1992. “Caldera” formation by unusually Geothermal Research 17:1Ð29. large phreatomagmatic eruptions through permafrost in arctic SIGURDSSON, H., and SPARKS, S. 1978. Rifting episode in Alaska. Eos 73:636. North Iceland in 1874Ð1875 and the eruptions of Askja and BEGÉT, J., MASON, O., and ANDERSON, P. 1991. Age and Sveinagja. Bulletin of Volcanology 41:149Ð167. extent of the ca. 3400 BP Aniakchak tephra. The Holocene SMITH, R.L., and BAILEY, R.A. 1968. Resurgent cauldrons. 2:117Ð123. Geological Society of America Memoir 116:613Ð662. CARR, M.H., and GREELEY, R. 1980. Volcanic features of WALKER, G.P.L. 1984. Downsag Calderas, ring faults, caldera Hawaii. , D.C.: National Aeronautics and Space sizes, and incremental caldera growth. Journal of Geophysical Administration. 211 p. Research 89:8407Ð8416. CAS, R.A.F., and WRIGHT, J.V. 1987. Volcanic successions: WILLIAMS, H. 1941. Calderas and their origin. Bulletin of the Modern and ancient. London: Allen and Unwin. 528 p. Department of Geological Sciences, University of , FUNICIELLO, R., PAROTTO, M., MARIOTTI, G., PREITE Berkeley 25:239Ð346. MARTINEZ, M., and TECCE, F. 1979. Geology, mineralogy, WILLIAMS, H., and McBIRNEY, A.R. 1979. Volcanology. San and stable isotope geochemistry of the Cesano geothermal field Francisco: Freeman and Cooper. 367 p. (Sabatini Mountains volcanic system, northern Latium, Italy). WILLIAMS, P.J., and SMITH, M.W. 1989. The frozen earth: Geothermics 8:44Ð73. Fundamentals of geocryology. Cambridge: Cambridge HOPKINS, D.M. 1959. History of Imuruk Lake, Seward Penin- University Press. 306 p. sula, Alaska. U.S. Geological Survey Bulletin 981. 54 p. WOHLETZ, K.H. 1986. Explosive magma-water interactions: ———. 1963. Geology of the Imuruk Lake area, Seward Peninsula, Thermodynamics, explosion mechanisms, and field studies. Alaska. U.S. Geological Survey Bulletin 1141-C. 21 p. Bulletin of Volcanology 48:245Ð 264. SEWARD PENINSULA MAARS ¥ 69

WOHLETZ, K.H., and HEIKEN, G. 1992. Volcanology and geo- YOKOYAMA, I.1958. Gravity survey on Kuttyaro Caldera lake. thermal energy. Berkeley: California University Press. 432 p. Journal of Physics of the Earth 17:75Ð79. WOOD, C.A., and KIENLE, J. 1990. Volcanoes of North America. ———. 1960. The subsurface structure of Oshima volcano, Izu. Cambridge: Cambridge University Press. 354 p. Journal of Physics of the Earth 18:55Ð68.